A Rethink of THD/THD+N (or why both tell us almost nothing)

Discussion in 'Audio Science' started by ChaChaRealSmooth, Jul 6, 2020.

  1. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    7,470
    Trophy Points:
    113
    Location:
    Winnipeg
    New dumb idea: generate tone(s) based on a piece of gear's distortion profile. Use that for evaluation and purchasing decisions rather than listening to the equipment itself.
     
  2. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    You do not need expensive equipment to test a new weighting algorithm. It can be done with the graphs containing FFT presented in the measurements posted, a scale or calipers, a pencil, piece of paper and a spreadsheet, calculator or even a slide rule.

    Example:
    1. print out distortion graph for amplifier of interest (FFT does not have any weighting applied)
    2. determine dB below 0 for each harmonic desired
    3. remember to convert dB back to linear before performing weighting calculations
    4. apply new weighting algorithm and calculate new metric
    5. convert back into dB or % as desired
    6. Read impressions on amplifier of interest. Best information is usually in the first page or two of impressions post.
    7. Correlate new metric to impressions
    8. Repeat for all amplifiers of interest.

    This only requires time and effort on your part, if you are interested.

    example graph for harmonics 2 through 6 (green circles on right channel)
    20201012 Bachelor GE5654W A04 THD+N  THD  nth-HD FFT 300R - 4+HD+N with 60Hz 0dBu annotated.png


    Now if you were intending that I do the following:
    1. spend 20 to 40 hours coding in a new algorithm into the analyzer, including testing new algorithm measurement accuracy
    2. remeasure a number of amplifiers, requiring 2 to 4 hours for each amplifier for a single measurement (remember amps should be thermal equilibrated before measurement)
    3. gather impressions information
    4. spend time making correlations
    5. post results for instant consumption

    Well, that would be annoying. I've already put in significant time on the 4+HD+N, which I believe is a useful metric, one that may be frequency swept for creating amplifier profile graphs with this metric. I am already on record as abhorring simplification of complex behavior to simple figure of merit numbers. There is another site on the internet that believes in that bullshit idea.
     
  3. mkozlows

    mkozlows Friend

    Pyrate
    Joined:
    Oct 6, 2015
    Likes Received:
    512
    Trophy Points:
    93
    Crutchfield does that for speakers.
     
  4. Ox Cart

    Ox Cart Facebook Friend

    Joined:
    Mar 13, 2018
    Likes Received:
    202
    Trophy Points:
    33
    Location:
    Austin, TX
    I find crap factor compelling, but in this specific example even the worse amp's crap factor is already -103dB. Is that really going to be noticeable on real world listening?
     
  5. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    Even the crap factor is still an over simplification. Look at the blue lines in the relevant post and note the greater downward slope of the preferred amp over the slight downward slope of the perceived lesser amp. Also remember to include the low NFB vs high NFB design criteria for the two amps. It is easy to achieve low distortion with high NFB. Much harder to achieve with low to no NFB.
     
  6. Ox Cart

    Ox Cart Facebook Friend

    Joined:
    Mar 13, 2018
    Likes Received:
    202
    Trophy Points:
    33
    Location:
    Austin, TX
    Yep, the distortion profile is different. I'm just trying to figure out the hypothesis here.

    I can buy the idea that some (maybe many) prefer amplifiers that manifest audible distortion following a decending amplitude profile. I'll also buy the idea that audible crap factor is a demerit. What I can't get my head around is that the distortion profile of an amp with overall inaudible distortion should matter. Is there something I'm missing?
     
  7. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,653
    Trophy Points:
    113
    Location:
    On planet
    The basilar membrane can operate over a 120 dB dynamic range. It also perceives sound differently inside a critical bandwidth than outside. For a give effective rectangular bandwidth (ERB) the operating dynamic range is about 30 dB. But different ERB may be operating at very different nominal levels. It is possible for the basilar membrane to detect 80 ~ 100 dB lower in one ERB far from another ERB. We can thank Bell Labs for a lot of work in discovering these principles. One of my mentors worked there for 26 yrs.
     
  8. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    • Like Like x 1
    • Dislike Dislike x 1
    • List
    Last edited: Jun 28, 2021
  9. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    It shouldn't.

    SINAD, THD+N, or even THD itself with the N were never meant as sound quality or transparency evaluators. They are really just numbers based from a simple steady state sine wave measurement, intended to assess one single aspect of the technical performance of an amp. In the late 1960s and early 1970s, a good number to shoot for was 0.01% THD for 1kHz into an expected load.

    Most of the great audio designers are dead or retired. I've had the pleasure of meeting many of the big names (and lesser known ones behind the scenes) on the account of where I used to live in CA (the San Fernando area used to be the Audio Mecca). I can tell you they just don't look at 1kHz THD, but also many other factors. There is a reason why my new Triple-Sine THD measurements are done with a bandwidth of 90kHz, even though we can't hear that high up.

    There is a lot of secret knowledge with respect to measurements on how something may sound which I am rediscovering today. I am beginning to sense that distortion harmonic distribution matters more than ultimate THD, especially when THD is good enough past a certain threshold. It's weird, but what I suspect is what are seeing are indicators of transient behavior which cannot be measured well (that is processed mathematically with error so we can visualize them properly - at least not without quantum computers).
     
    • Like Like x 5
    • Agreed, ditto, +1 Agreed, ditto, +1 x 1
    • List
  10. gepardcv

    gepardcv Almost "Made"

    Contributor
    Joined:
    Nov 20, 2015
    Likes Received:
    408
    Trophy Points:
    63
    Location:
    Terra, Sol System
  11. yotacowboy

    yotacowboy McRibs Kind of Guy

    Pyrate Contributor
    Joined:
    Feb 23, 2016
    Likes Received:
    10,697
    Trophy Points:
    113
    Location:
    NOVA
    Home Page:
    FYI:

    https://www.superbestaudiofriends.o...on-pass-diy-h2-injector-for-loaner-tour.8660/
     
  12. lemon07r

    lemon07r New

    Joined:
    Oct 1, 2021
    Likes Received:
    12
    Trophy Points:
    3
    Location:
    Canada
    I really want to know for myself as well, I really cant stand them. I liked my starfields more than my dusks cause of the weird BA sound it had.
     
  13. uncola

    uncola Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    596
    Trophy Points:
    93
    Location:
    Oahu, Hawaii
    "subjectivist objectivists" I feel like is basically accepting the value of measurements but also accepting fringe audio science hypotheses but only where actual research by smart experienced people is done.. and ignoring snake oil stuff that only exists to increase profit margins based on zero testing or research.

    pedja seems like he fits in with that
    https://www.monoandstereo.com/2008/04/interview-with-pedja-rogic-of-audial.html
     
  14. Elmer Danilovich

    Elmer Danilovich MOT:Earmen, HeadAmp, Bricasti; AKA:MShenay

    Contributor
    Joined:
    May 8, 2018
    Likes Received:
    346
    Trophy Points:
    63
    Location:
    In the Oak Forest
    I have to agree, that was an outstanding read and hope we will see even more manufacturers whose line of thought are similar to his
     
  15. wynpalmer

    wynpalmer Phono Pre Architect

    Pyrate
    Joined:
    Feb 23, 2023
    Likes Received:
    4
    Trophy Points:
    32
    Location:
    Greensboro, NC
    I'd be very interested in any comments concerning the following white paper.
    whitepaper (2).pdf (dropbox.com)
    It was intended as a superficial guide to the philosophy behind a phono stage design but seems to embody many of the criteria discussed in this thread.
    Criticism would be most welcome.
     
  16. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    I love these older books with formulas everywhere
     
  17. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,779
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Almost everything known in audio has been around for decades. There was a SINAD craze with opamps in the 70s and 80s (thanks to Doug Self, an earlier iteration of Amir, except Doug Self actually knew how to design stuff - I have his book). This is why the Ampzilla II and II used opamps. People started getting into specs, i.e. THD, even though Andy Hefley was the designer (per Craig / Eddie Current, one of the most brilliant solid-state discrete amp designer he's known). Andy has actually been to a few SBAF meets. Then of course another coming of the SINAD craze with NWAVGUY, which eventually died down, until Amir revived it yet again, but with better equipment that can measure noise on the molecular level, i.e. APx555. Thankfully, the SINAD craze is dying down again.
     
    • Like Like x 7
    • Agreed, ditto, +1 Agreed, ditto, +1 x 2
    • List
  18. CEE TEE

    CEE TEE MOT: NITSCH

    Pyrate IEMW
    Joined:
    Sep 25, 2015
    Likes Received:
    3,724
    Trophy Points:
    93
    Location:
    California Bay Area
    Home Page:
    Andy Hefley…also made the Grateful Dead Wall of Sound bass amps for Phil Lesh:
    9CD433F5-2327-46FE-8F0D-7A70D1C8E18E.jpeg
     
  19. penguins

    penguins Friend, formerly known as fp627

    Pyrate
    Joined:
    Sep 26, 2018
    Likes Received:
    3,451
    Trophy Points:
    113
    Location:
    SouthernCA
    I've had brief discussions with @ChaChaRealSmooth as well as a few others here at meets or whatever and I don't think anyone I've talked to before came up with an answer but just want to throw this out there:

    I don't think THD or THD+N by themselves are useless. Nor is a FR plot, CSD plot, most of @atomicbob 's measurements, decay plots occasionally posted by @purr1n , etc. However, these aren't so much wrong as they are varying degrees of incomplete, depending on the use case. Not because bob or purrin don't know what they're doing, but because existing equipment probably can't even characterize the other "things".

    My own theory has always been that the current "measurement suite" is not complete and on top of that, the measurements aren't necessarily geared towards telling us "how will this gear sound", but rather just measure one specific aspect of sound. While I have some conjectures and speculations about what other measurements could help tell us "how gear will sound", I will also be the first to say that these extra measurements, plots, or characteristics I'm guessing about still wouldn't complete the puzzle.

    On top of that, even if we had a magic wand that gave us a "complete measurement suite" that would tell us how something would sound, I would say these measurements still wouldn't tell us if any one of us individually would necessarily like how something would sound per these measurements as 1) they don't take personal differences in hearing into account 2) they won't account for measurable differences in our hearing that an audiologist can measure, or 3) account for individual hearing loss or imperfections.

    Practically speaking, if we're just talking audio, I'd rather just listen to something for a few seconds and then listen more as needed. But, now that I think about it, I almost wonder why studies haven't been done on this. Not just for audiophiles or audio sales and marketing, but perhaps for say academia, medical research, restoration of damaged hearing, etc.
     
    Last edited: Feb 27, 2023
  20. Donald North

    Donald North Friend

    Pyrate
    Joined:
    Oct 4, 2015
    Likes Received:
    932
    Trophy Points:
    93
    It’s always funny to me how many people focus on the measured attributes of amplifiers like distortion and phase/square wave response yet ignore the measured performance of loudspeakers using the same tests. Held to the same standard, >99% of speakers would be deemed junk/garbage.
     
    • Like Like x 6
    • Agreed, ditto, +1 Agreed, ditto, +1 x 3
    • Epic Epic x 1
    • List

Share This Page